Download Reversible lithium uptake by CoP3 at low potential: role of the anion

Electrochemistry Communications 4 (2002) 516–520
www.elsevier.com/locate/elecom
Reversible lithium uptake by CoP3 at low potential: role of the anion
V. Pralong, D.C.S. Souza, K.T. Leung, L.F. Nazar
*
Department of Chemistry, University of Waterloo, 200 University Ave. W., Waterloo, Ont., Canada N2L 3G1
Received 25 April 2002; accepted 3 May 2002
Abstract
Although various transition metal compounds in Group V (nitrides and antimonides) have been reported to act as low potential
Li insertion hosts, the phosphides have remained unexplored to date. We show here that lithium uptake and extraction in the metal
phosphide, CoP3 , provides a reversible capacity of 400 mAh/g at an average potential of 0.9 V vs Li=Liþ , via a novel mechanism.
The latter was revealed using a combination of X-ray diffraction, electron microscopy, and potentiodynamic and galvanostatic
intermittent measurements coupled with X-ray photoelectron spectroscopy (XPS). Initial uptake of Li forms highly dispersed cobalt
clusters embedded in a matrix of Li3 P; extraction of Li from this ion-conductive matrix on charge yields nano-particles of LiP, with
little change evident in the oxidation state of the Co site. This shows that contrary to the case of metal oxides here the anion plays the
major role in reduction and oxidation. We expect this is a general phenomenon for phosphides, and to be of fundamental interest
and future importance in the search for new negative electrode materials. Ó 2002 Published by Elsevier Science B.V.
Keywords: Cobalt phosphide; Anode; Lithium-ion battery; Transition metal pnictides; XPS; TEM
1. Introduction
The development and study of new negative electrode
materials for rechargeable lithium batteries remains a
major topic in the area of the energy storage. Alternatives to commercial carbonaceous materials are eagerly
sought to overcome the limitations of low gravimetric
and volumetric capacity and safety concerns that these
materials exhibit, and to match the higher capacities of
new emerging cathode materials. Meeting the requirements of reversibility and low cost is difficult, however.
Explorations of transition metal oxides as anode materials, first reported by Idota et al. [1], Nazar and coworkers [2] and by Tarascon and coworkers [3],
developed new concepts concerning the mechanism of Li
‘‘insertion’’. For example, in bulk MoO3 , Li uptake
drives the reduction of the metal to form a nano-phase
matrix of Li2 O, and metal suboxide particles: on charge,
Mo-oxide is regenerated, although ‘‘crystalline MoO3 ’’
is not fully achieved. Similarly, reduction of nano-phase
*
Corresponding author. Tel.: +1-519-888-4637; fax: +1-519-7460435.
E-mail address: [email protected] (L.F. Nazar).
1388-2481/02/$ - see front matter Ó 2002 Published by Elsevier Science B.V.
PII: S 1 3 8 8 - 2 4 8 1 ( 0 2 ) 0 0 3 6 3 - 6
CoO produces Co nano-particles in a matrix of Li2 O;
and nano-CoO is regenerated on charge [4]. It has subsequently been revealed that the latter reaction can be
thought of as a displacement reaction of the Co2þ ions
within the oxide matrix for Liþ on discharge, with the
reverse occurring on charge [5]. In these relatively
‘‘ionic’’ systems, reduction is driven by the difference in
thermodynamic free energy between the oxidized and
reduced states of the metal.
Recent interest has also been piqued by reports of Li
uptake in transition metal compounds of Group V, such
as nitrides [6] and antimonides [7], which show a lower
intercalation potential compared to the respective oxides. This results from the lower formal oxidation state
of the metal, and strong covalent character of the Mpnictogen bond, leading to high lying mixed anion-metal
bands, and a high degree of electron delocalization [8].
Where a fully lithiated end member exists, conversion
between two crystalline phases can occur on lithium
insertion, i.e., MnP4 $ Li7 MnP4 [9]. Here, we report a
completely different reversible Li uptake in metal
phosphides exemplified by CoP3 . We have found unique
behavior in this system: namely the redox behavior is
based on shuttling between two different phosphorus
oxidation states.
V. Pralong et al. / Electrochemistry Communications 4 (2002) 516–520
517
2. Experimental
3. Results and discussion
Synthesis of the cobalt phosphide was performed by
heating stoichiometric amounts of metallic cobalt and
red phosphorus at 650 °C for 24 h, in a stainless steel
tube sealed under argon. Powder diffraction patterns
were obtained using a Siemens D500 diffractometer
equipped with a diffracted beam monochromator (CuKa
radiation). The starting material, CoP3 was a pure
skutterudite phase, as confirmed by Rietveld refinement
of its XRD pattern that yielded values (cubic space
) as previously reported [10].
group IM3; a ¼ 7:7082 A
The structure consists of CoP6 octahedra corner-shared
to form P4 rings (Fig. 1, inset). The Li uptake properties
of the material were evaluated in Swagelok-type cells vs
the Li=Liþ couple using lithium as the anode. Mixtures
of the active material and 40 wt% carbon black were cast
on nickel current collectors. The cells were assembled in
an argon filled dry-box using a 1 M solution of LiPF6 in
1:1 EC/DMC as the electrolyte, and examined using a
MacPile controller (Biologic S.A., Claix, France).
X-ray photoelectron spectrocopy (XPS) analysis was
performed using a VG Scientific XPS Microprobe ESCALab 250 (with focused monochromatized AlKa radiation of energy 1486.6 eV, with a chamber pressure
<1010 mbar) on an irradiated area of 0:4 1 mm2 . The
cycled samples were opened in the dry-box, washed with
acetonitrile and transferred to the XPS analysis chamber
through a sample load-lock system to prevent air exposure. Charge compensation was employed during data
collection using a combination of a low energy flood gun
and low energy ions (care was taken to avoid charging).
These samples were also analyzed by means of transmission electron microscopy (TEM, Philips CM12).
The voltage vs composition curves for the insertion of
Li into CoP3 , acquired at a rate of 1Liþ per 10 h are
shown in Fig. 1. The first discharge process is characterized by a flat plateau corresponding to the uptake of 9
Li (after correction for the carbon contribution; the
initial feature in the curve at 0.8 V corresponds to the
SEI formation with carbon). On charge, 6 Li are extracted, leading to a reversible capacity of 1000 mAh/g
of CoP3 , that fades to 600 mAh/g after 10 cycles and
stabilizes at > 400 mAh=g. By means of potentiodynamic cycling with galvanic acceleration, a bell shaped
current response associated with a constant voltage
(0.35 V; not shown here) during the first discharge shows
that the plateau corresponds to a simple two-phase decomposition. We see no evidence for the formation of
intermediate ternary phases ‘‘Lix CoPy ; these are also not
expected since only a lithium-poor/metal-rich phase,
LiCo6 P4 , exists in the Li–Co–P phase diagram.
The reaction of CoP3 with 9 Liþ that we propose on
discharge: (CoP3 þ 9Liþ þ 9e ! 3Li3 P þ \Co") is
comparable to behavior reported by Tirado and coworkers [7] for CoSb3 , which has been proposed to
transform into ðLi3 Sb þ CoÞ on Li uptake. In the case of
CoP3 , the XRD pattern product at the end of the first
discharge (Fig. 2) clearly shows the reflections of the
Li3 P phase but not those of metallic cobalt. Any Co
particles must be of nano-sized dimensions (well below
the limit of detection in both XRD and TEM – see later).
The charge process, corresponding to the extraction
of 6 Li, and subsequent cycling behavior is less obvious.
The material becomes quasi-amorphous on oxidation,
Fig. 1. Potential/composition curve on the first galvanostatic cycling of
CoP3 –Li=Liþ cells performed at a nominal C/10 rate in the 1.7–0.2 V
potential window. Inset: Projection of the CoP6 octahedra along the
[0 0 1] direction in the structure of CoP3 :
Fig. 2. Ex situ XRD patterns of CoP3 at different cycling stages: (a)
before cycling; (b) discharge at 0.1 V after 1 cycle, Miller indices (*)
mark the position of Li3 P; (c) charge at 1.7 V after 1 cycle, reflections
mark the positions expected for LiP. The reflection of the cell holder
due to Al are indicated; ¼ unidentified phase arising from reaction
with the electrolyte.
518
V. Pralong et al. / Electrochemistry Communications 4 (2002) 516–520
as shown by X-ray diffraction and only very weak,
broad reflections are evident even after data collection
for extended periods (Fig. 2) [11]. An important question is whether the cationic or the anionic network plays
the major role in the electrochemical redox process.
That is, does oxidation result in re-formation of highly
dispersed CoP3 (as in the case of the oxide, CoO) or is
the process more complex? To more closely investigate
the lithium-driven structural and morphological changes, we examined the CoP3 electrode at various stages of
reduction and oxidation processes by means of transmission electron microscopy and X-ray photoelectron
spectroscopy.
The TEM micrograph of the starting material (Fig.
3(a)) shows it consists of particles of dimensions ranging
from 200 to 100 nm; as expected, selected area electron
diffraction (SAED) patterns taken down the [1 1 1]* zone
axis (Fig. 3(a), right) are consistent with the presence of
well-crystallized particles in accord with the XRD results. When the CoP3 is fully reduced (Fig. 3(b)), we
observed a complete change of the texture compared to
the starting material. The bright field image reveals the
(a)
(b)
(c)
Fig. 3. TEM images and corresponding SAED patterns on the right, of
the (a) starting material CoP3 , (b) fully discharged state, with the reflections of Li3 P indicated; (c) fully charged state showing the reflections of LiP – the inset shows the ‘‘halo’’ SAED image taken with a
normal diaphragm aperture, similar to that in (b).
particles embedded within the
presence of 10–20 A
crystallite, and preservation of the overall shape of the
starting particle. In this pseudomorphic reaction, a
partially amorphous material is formed, as confirmed by
the SAED pattern that consists of broad, weak rings.
The reflections were indexed to Li3 P (Table 1) in accord
with the XRD patttern. Analysis of the broad ring
pattern did not reveal any interplanar spacings assignable to cobalt crystallites (Table 1 and Fig. 3(b), right),
indicating that the metal must be very highly dispersed.
It is also possible that is weakly bonded to the matrix as
highly metal-rich phosphide clusters. Particle dimen would not be observable by SAED due
sions below 10 A
to extreme broadening.
On charge, the bright field image shows the pseudoamorphous character of the active material is preserved
although the particle size is more than halved (Fig. 3(c)).
Nonetheless, the weak, broad reflections observed in the
XRD pattern are resolvable, and identifiable by SAED
in the TEM owing to the smaller crystallite size that can
be probed by this method. The crystallites yielded a very
broadened SAED pattern consistent with their reduced
dimensions (Fig. 3(c), right, inset) in which only a few
concentric diffuse ‘‘halo’’ rings were visible, necessitating
employment of a larger diffraction diaphragm to improve the image quality 1; 2 [12,13]. The resulting SAED
pattern (Fig. 3(c), right) was best indexed to LiP (Table
1), but not to CoP3 . 3 As no lines can be ascribed to
Li3 P, it has been consumed during oxidation (Table 1).
Hence, the electrochemical process appears best characterized as a redox reaction between Li3 P and LiP.
Evidence for the reversibility, and nature of this intriguing process was supplanted by XPS measurements
that provide a direct ‘‘finger print’’ of the sample. The
binding energy is characteristic of the elements present
and their chemical states. Unfortunately, the resolvable
Co2p XPS region cannot be used to elucidate the exact
role of the cobalt, since metallic and covalent cobalt 2p
ð1=2; 3=2Þ states are at almost the same energy level;
furthermore the binding energy threshold does not
1
The diffraction diaphragm was increased by a factor of 10 to
increase the data sampling area: this explains the brightness of the
SAED image associated with a quasi-amorphous phase in the
relatively featureless XRD pattern. This technique is commonly used
for ‘‘amorphous’’ silicon, for example, where similar effects are seen for
nano-sized particles that do not show discrete features in their XRD
pattern [12].
2
The co-existence of an ‘‘amorphous’’ XRD pattern, and an
indexable ring pattern in the SAED using a standard aperture has
also been observed; see [12].
3
Reflection positions matched those of LiP: P21/c, a ¼ 5:582(1)
b ¼ 4:94(1) c ¼ 10:255(2), b ¼ 118:150, JCPDF no. 83–1575. Small
deviations may result from the presence of defect-laden LiP, owing to
its nano-dimensions and method of formation. Note that the two
strongest lines of CoP3 : (2 0 0) and (2 2 0), do not match the observed
lines, Table 1.
V. Pralong et al. / Electrochemistry Communications 4 (2002) 516–520
519
Table 1
List of expected and observed d-spacings from XRD and SAED patterns for the products after discharge and charge
Li3 P, calculated
)
reflections (A
d-spacing, SAED
)
pattern (A
d-spacing, XRD
)
pattern (A
(a) discharged state
3.69 (1 0 1)
2.14 (1 1 0)
1.86 (1 1 2)
1.42 (2 1 0)
1.22 (3 0 0)
3.69
2.14
1.85
1.40
1.23
3.5
2.1
1.8
–
–
LiP, calculated
)
d-reflections (A
d-spacing, SAED
)
pattern (A
Approximate
)
d-spacing (XRD) (A
CoP3 , calculated reflections
)
near observed (A
(b) charged state
3.34 (0 1 2)
2.94 (1 1 1)
2.46 (2 0 0)
2.27 ()1 1 4)
3.40
2.94
2.46
2.21
3.4
3.0
–
–
3.85
2.72
2.44
2.22
change significantly from the pristine material upon reduction or oxidation (Co2p1=2;3=2 at 793.5/778.4 eV in
CoP3 ; 794.0/779.0 eV on discharge; at 793.7/778.7 eV on
charge). Examination of the overlapping Co3p and Li1s
region is more revealing, however: Fig. 4 shows the
spectra of the fully charged and discharged electrodes,
after ion sputtering to clean the surface. In both spectra,
the broad feature (P1) at 60 eV is assigned to the Co3p
ð1=2; 3=2Þ states, by comparison with reference data
[14]. This peak should, of course, remain unchanged in
accord with the invariance of the Co3p states (see
above). The source of the change in the spectra therefore
must lie in the Li 1s states. In both spectra, the signature
at 56 eV (P2) corresponds to Liþ from the residual
electrolyte (LiPF6 ) [14]. In the spectrum of the discharged material (Fig. 4(a)), this peak is overlapped with
(a)
P1
P2,3
(b)
P1
P3
P2
2.05 (1 1 1)
1.78 (2 0 0)
(2 0 0)
(2 2 0)
(0 1 3)
(2 2 2)
another at 56.87 eV, but is distinguishable in a curve
deconvolution. This peak (P3) is assigned to the Li 1s
state in Li3 P, in good accordance with the literature
value of 56.9 eV [14]. It shifts to a higher binding energy
(58.2 eV) on charge (Fig 4(b)), indicating the phosphorous oxidation state decreases. This peak is assigned to
LiP by comparison with its XPS spectrum (binding energy: 59.03 eV). 4 A simplified mechanism accounting
for the initial reversible capacity of 6 Li can therefore be
written as follows:
First discharge process :
CoP3 þ 9Liþ þ 9e
! 3Li3 P þ \Co"
Subsequent cycling :
3Li3 P () 3LiP þ 6Liþ þ 6e
In this system, the relatively high electronic conductivity of CoP3 enhances initial electronic transport
within the electrode material and facilitates the decomposition process. After discharge, the high ionic conductivity of Li3 P of 103 S/cm (that has spurred its
investigation as a solid electrolyte [15]) promotes Li ion
transfer within the composite matrix. The role of the
cobalt in the redox process (if any) is still unclear; since
the cobalt is in a highly dispersed state in both the discharged and charged materials which preclude investigation of its nature, we can only conclude that it acts as
an ‘‘active spectator’’ after initial discharge as bulk
metallic cobalt is not formed. This is also in accordance
with the high levels of carbon that are needed in the
compound electrode to promote electronic conductivity.
The high degree of Co dispersion and bonding to the Li–
4
Fig. 4. XPS Lithium-1s spectra of Lix CoP3 in its (a) fully discharged
and (b) fully charged states.
Co metal, calculated reflections
)
near observed (A
This spectrum was collected on the reference material, LiP,
synthesized from heating (Li0 þ red P) at 600 °C for 24 h in a sealed
stainless steel tube. Note that the binding energy of bulk LiP and that
prepared in situ differ by about 0.2 eV, consistent with the shift
expected for nano-particles in the quantum size regime.
520
V. Pralong et al. / Electrochemistry Communications 4 (2002) 516–520
P matrix may nonetheless enhance Li transport by grain
boundary effects.
The fact that neither CoP3 nor an amorphous form of
lithium cobalt phosphide are regenerated on charge sets
this mechanism apart from transition metal oxides,
where partial or complete recuperation of the oxide
structure occurs. This can be understood since the reformation of CoP3 nano-crystallites is unlikely in the
more covalently bonded phosphides, owing to the
thermodynamics of the Li–Co–P system which should
favor LiP formation on charge as a result of its high
lattice energy. Nonetheless, the fully reduced material
can be considered as a highly dispersed, phase-separated
metastable analog of ‘‘Li6 CoP3 ’’. In this respect, the
process is different from Li uptake in phosphides where
ternary lithiated phases do exist, such as MnP4 , In this
case, reversible conversion to crystalline Li7 MnP4 results [9]. Both processes, however, are initiated by electron uptake that primarily occurs at the anion centre,
giving rise to scission of the P–P bonds.
4. Conclusions
This work highlights the importance of the anion in
the electrochemical process, proving that the redox active centers are not necessarily cationic in transition
metal compounds but can be based almost uniquely on
the anionic network. In this regard, it is worthwhile
recalling that the anion alone also acts as the redox
center in the well-known Li–sulfur battery. This cell is
based on reactions of Li and sulfur that reversibly yield
Li2 S. Development of those cells has been hampered by
the high ionic and electronic resistivity of sulfur, solubility of polysulfide products, and other factors [16].
Investigations of transition metal pnictides as anodes for
Li-ion batteries may re-open new opportunities in this
fascinating area. Such studies would also focus on the
aspect of microstructural control that would lead to
improve cycling stability in these materials, work that is
currently in progress in our lab.
Acknowledgements
The authors are grateful to B. Beaudoin (Amiens),
M.M. Treacy (NEC labs, Princeton), and D. Perovic (U
of Toronto), and for insightful discussions on TEM.
D.K. Fujita and Dr. Enke are thanked for their help.
D.C. Souza thanks CAPES/Brazil for their financial
support in the form of a scholarship, and LFN gratefully acknowledges the support of NSERC for funding
this work.
References
[1] Y. Idota, US Patent, 5478671 (1995);
H. Tomyama, Jap. Patent, 07-029608 (1995);
C. Sigala, D. Guyomard, Yves Piffard, M. Tournoux, C. R. Acad.
Sci. Paris 320 (1995) 523.
[2] F. Leroux, G.R. Ouvrard, W.P. Power, L.F. Nazar, Electrochem.
Solid State Lett. 1 (1998) 255;
F. Leroux, L.F. Nazar, New Mater. Batteries Fuel Cells II 55
(2000) 173.
[3] S. Denis, E. Baudrin, F. Orsini, G. Ouvrard, M. Touboul, J.-M.
Tarascon, J. Power Sources 81 (1999) 79.
[4] P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J.M. Tarascon,
Nature 407 (2000) 496.
[5] M.N. Obrovac, R.A. Dunlap, R.J. Sanderson, J.R. Dahn,
J. Electrochem. Soc. 148 (2001) A576.
[6] M. Nishijima, T. Kagohashi, M. Imanishi, Y. Takeda, O.
Yamamoto, S. Kondo, Solid State Ionics 83 (1996) 107;
T. Shodai, S. Okada, S. Tobishima, J. Yamaki, J. Solid State
Ionics 86–88 (1996) 785;
Y. Takeda, M. Nishijima, M. Yamahata, K. Takeda, N. Imanishi,
O. Yamamoto, Solid State Ionics 130 (2000) 61.
[7] R. Alcantara, F.J. Fernandez-Madrigal, P. Lavela, J.L. Tirado,
J.-C. Jumas, J. Olivier-Fourcade, J. Mater. Chem. 9 (1999)
2517.
[8] M. Lefebvre-Devos, M. Lassalle, X. Wallart, J. Olivier-Fourcade,
L. Monconduit, J.-C. Jumas, Phys. Rev. B 63 (2001) 125110;
L. Monconduit, M. Tillard-Charbonnel, C. Belin, J. Solid State
Chem. 156 (2001) 37.
[9] D.C.S. Souza, V. Pralong, A.J. Jacobson, L.F. Nazar, Science 296
(2002) (in press).
[10] W. Biltz, M. Heimbrecht, Z. Anorg. Allgem. Chem. 241 (1939)
238.
[11] Data were collected on a Siemens D500 equipped with a graphite
monochromator using Cu–Ka radiation, and a step size of 0.05°
with a counting time of 30 s on each step.
[12] M.M. Treacy, NEC, Princeton Labs, Personal communication,
2002.
[13] S. Imamura, S. Ishida, H. Tarumoto, Y. Saito, T. Ito, J. Chem.
Soc. Faraday Trans. 89 (1993) 757.
[14] J. Chastain (Ed.), Handbook of X-ray Photoelectron Spectroscopy, Perkin–Elmer Corporation, Eden Prairie, 1992.
[15] G. Nazri, Solid State Ionics 34 (1989) 97.
[16] D. Marmorstein, T.H. Yu, K.A. Striebel, F.R. McLarnon, J. Hou,
E.J. Cairns, J. Power Sources 89 (2000) 219.